Buggy Mat.at with 3 indexes (Bug #2233)

Added by NoRG Zotto over 12 years ago. Updated over 12 years ago.

Status:Cancelled Start date:2012-08-03
Priority:Normal Due date:
Assignee:- % Done:

0%
Category:core
Target version:-
Affected version: Operating System:
Difficulty: HW Platform:
Pull request:

Description

Such example code (simplified from cornerDetector_Demo.cpp) raises assert:

Mat myHarris_dst = Mat::zeros(Size(100, 100), CV_32FC(6) );
float lambda_1 = myHarris_dst.at<float>( 0, 0, 0 );

OpenCV Error: Assertion failed (elemSize() == (((((DataType<_Tp>::type) & ((512- 1) << 3)) >> 3) + 1) << ((((sizeof(size_t)/4+1)*16384|x3a50)>> ((DataType<_Tp>::type) & ((1 << 3) - 1))*2) & 3))) in unknown function, file C:\usr\local\opencv-249\include\opencv2/core/mat.hpp, line 593

Two-indexes "at" looks like not bugged. I'm on master branch, 2.42 also bugged. Maybe I just missed with syntax?

Associated revisions

Revision 7942494f
Added by Vadim Pisarevsky almost 11 years ago

Merge pull request #2233 from palindromoroz:BGSubtract_KNN_new_code

History

#1
Updated by Daniil Osokin over 12 years ago

I tried to reproduce that, but this two lines of code works as well as usual. No any asserts. Suppose, that you have assert by another reason.

#2
Updated by NoRG Zotto over 12 years ago

Maybe you run in release mode? CV_DbgAssert raises only in Debug mode, error located here

template<typename _Tp> inline _Tp& Mat::at(int i0, int i1, int i2)
{
    CV_DbgAssert( elemSize() == CV_ELEM_SIZE(DataType<_Tp>::type) ); /// <-  here raised
    return *(_Tp*)ptr(i0, i1, i2);
}

Or maybe your compiler removed this block when optimizing.

Please, try this.

    Mat myHarris_dst2 = Mat::zeros(Size(100, 100), CV_32FC(6) );
    myHarris_dst2.at<float>( 5, 5, 5 ) = 100.f;
    float lambda_1 = myHarris_dst2.at<float>( 5, 5, 5 );

Or best just try compile and run samples\cpp\tutorial_code\TrackingMotion\cornerDetector_Demo.cpp.
I compiling by MS VS2010 x86.

#3
Updated by Daniil Osokin over 12 years ago

at<i, j, k> works with 3D matrix, but you have 2D, so that's the reason. It's not a Bug, good place for that kind of question is http://answers.opencv.org/questions/, you can use it in future.

  • Target version deleted (3.0)
  • Assignee deleted (Vadim Pisarevsky)

#4
Updated by Vadim Pisarevsky over 12 years ago

to access n-th channel of element (x,y) of 100x100 6-channel matrix, you need to use

myHarris_dst.at<float>(y, x)[n] instead of myHarris_dst.at<float>(y, x, n).

  • Status changed from Open to Cancelled

Also available in: Atom PDF